实数编码化学反应优化的全局收敛性研究

作为一种新的自然计算方法,化学反应优化性能优越,适应性强,但其理论研究缺乏.针对上述问题,以实数编码化学反应优化(RCCRO)为对象,研究其收敛性和收敛速度.首先,建立RCCRO在连续时间上有限吸收的Markov链模型,并证明其为有限可吸收Markov链;然后,基于有限吸收Markov链证明RCCRO的收敛性;进而,对于采用不同初等反应组合的RCCRO,研究初等反应的有效性和算法全局收敛的必要条件;最后分析了RCCRO的收敛速度和首达时间.

On the global convergence of real-coded chemical reaction optimization

Chemical reaction optimization is a kind of new natural calculation methods due to its superior performance and strong adaptability, but it lacks theoretical analysis. Aiming at this problem, we study the convergence of the real-coded chemical reaction optimization as well as its convergence speed. First, set up a RCCRO Markov chain model in its continuous time and prove that it is finite and absorbable. Second, prove its convergence through this finite and absorbable RCCRO Markov chain. Third, by adopting different combinations of the elementary reactions, study its effectiveness and the necessary conditions of global convergence to this algorithm. Finally, analyze the convergence speed and the first arrive time of RCCRO.